Ask HN: Deploy ML Model as API

Looking to deploy some pre-trained Keras models into a simple but scalable RESTful API.

What would the easiest way of doing this be? And what would the cheapest way be? (Presuming these are different)

My current thoughts are to save Keras model as .h5 and combine with Flask and AWS EC2.

New to deployment and backend stuff so would be very appreciative for any help. Thanks so much!